Received: 16 January 2019 Revised: 19 February 2019 Accepted: 5 March 2019

DOI: 10.1111/obr.12857

NUTRITION

Impacts of carbohydrate‐restricted diets on micronutrient

intakes and status: A systematic review

Chaitong Churuangsuk | Daniel Griffiths | Michael E.J. Lean | Emilie Combet

Human Nutrition, School of Medicine

Dentistry and Nursing, College of Medical Summary
Veterinary and Life Sciences, University of A systematic review of published evidence on micronutrient intake/status with
Glasgow, Glasgow, UK
carbohydrate‐restricted diets (CRD) was conducted in Web of Science, Medline,
Correspondence Embase, Scopus, CENTRAL, and ClinicalTrials.gov up to October 2018. We identified
Dr Emilie Combet, Room 2.22, Level 2, New
Lister Building, Glasgow Royal Infirmary, 10‐16 10 studies: seven randomized controlled trials (RCTs) (“Atkins”‐style, n = 5; “Paleo-
Alexandra Parade, Glasgow, G31 2ER, UK. lithic” diets, n = 2), two Atkins‐style noncontrolled trials and one cross‐sectional
Email: emilie.combetaspray@glasgow.ac.uk
study. Prescribed carbohydrate varied 4% to 34% of energy intake. Only one noncon-
Funding information trolled trial prescribed multivitamin supplements. Dietary intakes/status were
Prince of Songkla University
reported over 2 to 104 weeks, with weight losses from 2 to 9 kg. No diagnoses of
deficiency were reported. Intakes of thiamine, folate, magnesium, calcium, iron, and
iodine all decreased significantly (−10% to −70% from baseline) with any CRD types.
Atkins diet trials (n = 6; 4%‐34%E carbohydrate) showed inconsistent changes in vita-
min A, E, and β‐carotene intakes, while a single “Paleolithic” diet trial (28%E
carbohydrate) reported increases in these micronutrients. One other “Paleolithic” diet
(30%E carbohydrate) reported a rise in moderate iodine deficiency from 15% to 73%
after 6 months. In conclusion, few studies have assessed the impacts of CRD on
micronutrients. Studies with different designs point towards reductions in several
vitamins and minerals, with potential risk of micronutrient inadequacies. Trial
reporting standards are expected to include analysis of micronutrient intake/status.
Micronutrients in foods and/or supplements should be considered when designing,
prescribing or following CRDs.

K E Y W OR D S

low‐carbohydrate diets, micronutrient, minerals, vitamins

1 | I N T RO D U CT I O N National Trends Survey (USA) of 5586 participants reported high

awareness of CHO‐restricted diet (86.6%) with 17% of responders
Carbohydrate (CHO)‐restricted diets have been promoted and used having tried this diet.1 In the United Kingdom, media reporting sug-
globally for weight management, and for treatment of diabetes, under gests that 7% of men and 10% of women are trying CHO‐restricted
either medical supervision or self‐guidance. The Health Information diets,2 similar to data from Finland.3
The Active “Low‐Carber” Forums online survey (n = 3134)
ABBREVIATIONS: CENTRAL, Cochrane Central Register of Controlled Trials; CHO, reported that only 50% of respondents had support from their
carbohydrate; DRV, dietary reference values; EAR, estimated average requirement; PRISMA,
doctors, while 6% reported that their doctors did not encourage them,
Preferred Reporting Items of Systematic reviews and Meta‐analyses; RBC, red blood cell;
RCT, randomized controlled trial; RNI, Reference Nutrient Intake despite achieving weight loss. Online support forums were the most

1132 © 2019 World Obesity Federation wileyonlinelibrary.com/journal/obr Obesity Reviews. 2019;20:1132–1147.

CHURUANGSUK ET AL. 1133

important source of information for 60% of the respondents, while or nonrandomized controlled trials), cohort, cross‐sectional or case‐
75% reported that government websites/publications were not control studies. CHO‐restricted diets included the Atkins diet, the
important to them.4 Zone diet, the “Paleolithic” diet, ketogenic diets, or a prescribed
Although a topic of scientific and public debates, extensive and CHO‐restricted diet as stated by the authors of the published studies.
rigorous evidence review finds no clear evidence from systematic Papers were excluded if the study included a cointervention (such as
reviews and meta‐analyses of randomized controlled trials (RCTs) for drug or exercise administered alongside CHO‐restricted diet), subjects
superiority from either high‐ or low‐CHO diets for weight control or 18 years old or younger, and did not report micronutrient intakes
5-8
for diabetes care. Most clinical guidelines allow flexibility over and/or status in each diet arm (for intervention studies).
CHO content, to suit personal preference, but the role of high‐CHO
foods as sources of micronutrients appears rather neglected in this
2.3 | Study selection and data extraction
debate.9,10 Very limited previous evidence has suggested that
micronutrient content and adequacy may be compromised.11
All records retrieved from each database were exported to EndNote
Restricting consumptions of CHO‐rich foods such as wholegrains,
(EndNote X8, Thomson Reuters, New York, USA). After duplicates
cereals, and fruits might reduce consumptions of B‐vitamins, minerals,
removal, two reviewers, CC and DG, independently screened titles,
and antioxidant nutrients, potentially risking deficiencies. The present
abstracts, and full‐texts against eligibility criteria. A third reviewer
systematic review has evaluated up‐to‐date published literature to
(EC) was consulted when there was disagreement between the two
assess the reported impacts of CHO‐restricted diet on micronutrient
reviewers. Data extraction was performed in the template created
intake and micronutrient status.
using Microsoft Excel by CC and DG independently, and disagree-
ments were solved by consensus and consultation with the third
reviewer (EC). The extracted data included authors, year, study design,
2 | METHODS
participant characteristics, sample size, details of intervention, dietary

This systematic review was conducted following a protocol registered intakes, duration of study, micronutrient intakes/status, and supple-

in PROSPERO, registration number CRD42018085483, and it has ment use.

been reported in accordance to the Preferred Reporting Items of Sys-

tematic Reviews and Meta‐analyses (PRISMA) guideline.12 2.4 | Risk of bias assessment

2.1 | Search and information source Two reviewers independently assessed the risk of bias among
included studies using the Cochrane Risk of Bias Tool, for RCTs, which
A systematic literature search was conducted in electronic databases comprises of six categories: random sequence generation, allocation
as following Web of Science Core Collection, Medline (OVID), concealment, blinding of participants and personnel, blinding of out-
EMBASE (OVID), Scopus, Cochrane Central Register of Controlled come assessment, incomplete outcome data, and selective reporting.
Trials (CENTRAL), and ClinicalTrials.gov for trial registry record until The nature of the trials required open intervention with no blinding
October 2017, for initial searching. Having established the methodo- of the trial participants or the investigators. For observational studies
logical approach, we updated search for Web of Science Core and noncontrolled trials, the quality assessment tools for observational
Collection, Medline (OVID), EMBASE (OVID), and Scopus databases cohort and cross‐sectional studies, and for before‐after (pre‐post)
in October 2018. We also searched for additional articles from refer- studies with no control group (National Institute of Health) were used
ence lists of included full‐texts and consulted experts in the field for respectively. Risk of bias assessments was conducted for micronutri-
relevant articles. ent intake outcome and micronutrient status outcome separately.
Search limit of English language and human studies were applied.
Search terms as free texts and MeSH terms related to CHO‐restricted
2.5 | Data synthesis
diet and micronutrient intakes or statuses were used, including the
following micronutrients found in CHO rich foods: thiamine, riboflavin, A narrative synthesis was implemented for the present systematic
niacin, folate, cobalamin, zinc, magnesium, selenium, and iron. Full review, as there were limited numbers and high heterogeneities of
search strategy for Medline is available online (Table S1). studies included. Actual intakes of micronutrients were translated into
percentages of recommended intakes according to the guideline of the
2.2 | Eligibility criteria country where the study was conducted, including the United
Kingdom, the United States, Australia, Sweden, and Iceland. Changes
Papers were included if they reported micronutrient intakes and/or from baseline intake of each micronutrient were calculated as percent-
status in adult participants (18 years or older) who have followed age for each time point of follow‐up. Among weight loss trials,
CHO‐restricted diet as either being assigned to CHO‐restricted correlation analyses between changes in body weight and reduced
diet intervention or being identified in observational studies. The intakes of each micronutrient were performed using Spearman's rank
study designs allowed any intervention studies (either randomized correlation.
1134 CHURUANGSUK ET AL.

3 | RESULTS 3.2 | Study characteristics

A total of 595 records was retrieved from literature searching and 62 Among clinical trials, the Atkins and Atkins‐style diets were used in
records from additional sources. After duplicates removal, 533 records five RCTs and two noncontrolled trials and the “Paleolithic” diet in
were screened on the basis of title and abstract by two researchers two RCTs. Durations of intervention ranged from 2 weeks to
independently, resulting in 52 full‐texts to be assessed against the 24 months (104 weeks). All trials with Atkins‐style diet (n = 7) were
inclusion and exclusion criteria. A total of 10 studies were eligible, conducted in adults with overweight and obesity, but neither
including seven RCTs, two noncontrolled clinical trials, and one established cardiovascular diseases nor diabetes mellitus. One
cross‐sectional study (Figure 1). Excluded full‐texts are shown in Atkins‐style diet trial included patients with high liver fat content.
Table S2, with reasons. Two studies with the “Paleolithic” diets were conducted in self‐
reported healthy women and postmenopausal women. Sample sizes
ranged from 10 to 293 participants (Table 1).
Eight clinical trials aimed primarily to assess the effects of
3.1 | Risk of bias and quality assessment CHO‐restricted diet for weight loss and body composition, either with
or without comparison to other weight loss diets (six Atkins‐style diets
The risk of bias of included RCTs is shown in Table 1 (including and two “Paleolithic” diets). Of these, one crossover RCT additionally
Tables S3 and S4 for detailed assessment) for micronutrient intakes assessed effects of Atkins diet on antioxidant vitamin status. One other
and status; all studies were of low risk of bias. The blinding of the noncontrolled trial assessed the effect of “isocaloric” CHO‐restricted
outcome assessors was judged as low risk of bias for all RCTs, as the diet on liver fat content and gut microbiota without body weight
outcomes are objective (both dietary intakes and plasma/urine bio- loss in patients with high liver fat content. There was no report of
markers of micronutrients). Two RCTs were judged as at high risk of clinically diagnosed micronutrient deficiency among the included
attrition bias due to substantial dropout rate (ie, greater than or equal trials. Only one trial prescribed multivitamins supplement to the
to 20%). Two noncontrolled trials of Atkins and Atkins‐style diets met participants, with one other trial reporting the number of participants
9/10 items of the quality assessment tool for pre‐post studies; and a who started taking supplements after the CHO‐restricted diet. The rest
cross‐sectional study met 6/8 items of the quality assessment tools of the included studies did not report on supplement use (Table 1).
for observational cohort and cross‐sectional studies. These three Four clinical trials provided diet books for participants to follow
studies were also judged as low risk of bias (Tables S5 and S6). without any further instructions, while three clinical trials conducted

FIGURE 1 Study selection flow diagram.

WoS, Web of Sciences; CENTRAL, Cochrane
Central Register of Controlled Trials; CRD,
carbohydrate‐restricted diet [Colour figure
can be viewed at wileyonlinelibrary.com]
TABLE 1 Characteristics of included studies (n = 10)a

Primary Outcome of Duration of Dietary Assessment Supplement Overall

Author/Country Study Design Study Population Age, BMI Studies Interventions Methods Use Risk of Bias
CHURUANGSUK

Miller 2003 Single‐arm 9 men and 9 women, Age 39.8 ± 8 y Weight loss and dietary 4 wk 3‐d dietary record Multivitamin Low
USA29 clinical trial overweight and obesity, BMI 36.4 ± 6.5 kg/m2 intake daily
ET AL.

without clinical CVD provided by

and diabetes. a research
team
Brehm 2003 Parallel RCT 53 women with obesity, Age 44.2 ± 7 y Weight loss 6 mo 3‐d weighed Not reported Low
USA30 without clinical CVD BMI 33.2 ± 1.8 kg/m2 food record
and diabetes.
(completers = 42)
Truby 2008 Parallel RCT 293 men and women, Age 40 ± 10 y Weight loss 6‐mo trial but dietary 7‐d estimated Not reported Low
UK27 overweight and obesity, BMI 32 ± 3 kg/m2 intake reported dietary record
without clinical CVD only over the first
and diabetes. 2 mo
(completers = 240)
Gardner 2010 Parallel RCT 311 women, overweight Age 42 ± 5 y Weight loss 12‐mo trial but 3 separate, Voluntarily Low
USA26 and obesity, without BMI 32 ± 4 kg/m2 dietary intake nonconsecutive, taking
clinical CVD reported only 24‐h dietary MTV/MTM,
and diabetes. over the first 8 wk recalls n = 3;
(completers = 291) calcium,
n=3
Johnston 2011 Crossover RCT 16 men, overweight and Age 55 ± 14 y Weight loss and 4 wk of CRD Research staff Not reported Low
UK25 obesity without clinical BMI 36 ± 6 kg/m2 antioxidant crossover with supplied all
CVD and diabetes. micronutrient 4 wk of MCD. foods and drinks
status
Bazzano 2014 Parallel RCT 148 men and women Age 45.8 + 9.9 y Weight loss 12 mo 2 d 24‐h dietary Not reported Low
USA31 without clinical CVD BMI 35.2 + 3.8 kg/m2 recalls
and diabetes.
(completers = 119)
Genoni 2016 Parallel RCT 42 healthy women Age 47 ± 13 y Body weight change, 4 wk 3‐d weighed Not reported Low
Australia16 (completers = 39) BMI 27 ± 4 kg/m2 CVD risk factors food record
Manousou 2018 Parallel RCT 70 women, Age 60 ± 6.3 y Weight loss and body 24 mo 4‐d estimated Not reported Low
Sweden17 postmenopausal, BMI 32.6 ± 3.4 kg/m2 composition dietary record
overweight and obesity
without clinical CVD
and diabetes.
(completers = 49)

(Continues)
1135
 1136 CHURUANGSUK ET AL.

counselling sessions, two short‐term trials (2‐4 weeks) provided all

Abbreviations: BMI, body mass index; CRD, carbohydrate‐restricted diet; CVD, cardiovascular disease; MCD, moderate carbohydrate diet; MTV/MTM, multivitamins/multimineral; N/A, not applicable; RCT, ran-
Risk of Bias
foods and drinks to their participants. A cross‐sectional study of
Overall
people voluntarily following CHO‐restricted diet did not report

Low

Low
on the sources of their CHO‐restricted diet information. Amount
of CHO contributing to total energy intake varied among studies
Not reported

Not reported
ranging from 4% to 34%. However, mean CHO intakes met
Supplement

prespecified CHO contents (adherence) in only 4/6 trials during

4 weeks to 3 months treatment time (Table 2).
Use
Dietary Assessment

3.3 | Changes in body weight and energy intake in

All food provided

the included studies

food record
3‐d weighed
Methods

Of eight weight loss trials, reported energy intakes per day decreased
by approximately −300 to −1000 kcal (median −656 kcal), (−1255 to
−4184 kJ; median −2745 kJ), and body weight changes ranged from
−3.2 to −8.5 kg over 2 weeks to 24 months. In contrast, a study of
“isocaloric” CHO‐restricted diet that aimed to maintain body weight
Interventions

reported an increase in reported energy intake of +880 kcal but

Duration of

demonstrated mean weight loss of −1.9 kg (Tables 3 and 4).

2 wk

N/A

3.4 | Micronutrient intakes

Primary Outcome of

intakes and CVD

Micronutrient intakes among the studies included were assessed by

To assess dietary

self‐reported dietary records (n = 6), and 24‐hour dietary recalls

risk factors
Liver fat loss

(n = 2), while two of the studies provided all foods and drinks to
Studies

their participants (Table 1). Tables 3 and 4 show amounts of micro-

nutrient intakes per day, percentage of recommended intake in the
country where the study was conducted, and the changes from
BMI 34.1 ± 1.2 kg/m2

baseline of each micronutrient in included studies. Numbers of stud-

BMI 30 ± 4 kg/m2
Age 53.7 ± 3.65 y

ies that reported micronutrients are as follows: thiamine (n = 6),

Age 43 ± 11 y

folate (n = 8), vitamin B12 (n = 4), vitamin C (n = 9), vitamin A

(n = 7), β‐carotene (n = 4), vitamin E (n = 6), calcium (n = 7),
Age, BMI

magnesium (n = 6), iron (n = 6), iodine (n = 2), zinc (n = 5), and sele-
nium (n = 4).
Among all weight loss trials (reduced energy intake; assessed by
10 adults (8 men) with high

dietary records and/or 24‐hour recalls), there were reductions (from

Values are presented as mean ± SD unless otherwise indicated.

baseline) in thiamine intake (−35 to −68%, n = 5), folate (−15 to

54 adults voluntarily
liver fat content
Study Population

−71%, n = 6), calcium (−13 to −63%, n = 5), magnesium (−9 to −54%,

following CRD

n = 4), iron (−13 to −57%, n = 4), and iodine (−47%, n = 1). All these
micronutrients were also lower than the country‐specific recom-
mended intakes and accounted for 35% to 80% of recommended
intake. One single‐arm trial with “isocaloric” intake (all food provided)
in patients with high liver fat showed a significant increase in thiamine,
Cross‐sectional
clinical trial
Study Design

but reductions in magnesium and iodine intakes, while no significant

Single‐arm

study

changes for folate, calcium, or iron (Tables 3 and 4). When comparing
to control diets, most of the RCTs showed that participants in CHO‐
domized controlled trial.
(Continued)

restricted diet group had lower intakes of thiamine, folate, calcium,

magnesium, iron, and iodine (Table 5).
Author/Country

Elidottir 2016

Approximately 30% increase in vitamin B12 intake was reported in

Sweden28
Mardinoglu

Iceland32

clinical trials of 4‐ to 8‐week Atkins diet compared with baseline,

2018
TABLE 1

twofold to sixfold higher than the recommended intake. Atkins diet

trials reported reduced vitamin C intake, by −3% to −71% from
a
TABLE 2 Mode of diet delivery, prescribed carbohydrate‐restricted diets, and macronutrient intakes in the studies includeda [Colour table can be viewed at wileyonlinelibrary.com]

Comparator
Author/Country Mode of Diet Delivery Prespecified CRD Baseline CRD (Intermediate) CRD (Last Time Point) (Last Time Point)
CHURUANGSUK

Miller 2003 A copy of Dr Atkins' <20 g CHO per day, Baseline Induction 2 wk OWL: 4 wk N/A
ET AL.

USA29 New Diet Revolution ad libitum for fat. E 2481 ± 723 kcal E 1400 ± 472 kcal E 1558 ± 490 kcal
book was provided. Add 5 g CHO per day (10.4 ± 3 MJ) (5.9 ± 2 MJ) (6.5 ± 2 MJ)
up to 30 g/d during CHO 43%, Pro 16% CHO 21 g (6%), CHO 24 g (6%),
ongoing weight loss Fat 41% Pro 29%,Fat 64% Pro 29%, Fat 64%
weeks.
Brehm 2003 Dietitians led group <20 g CHO per day, ad Baselineb 3 mob 6 mob LFD 6 mob
USA30 sessions for libitum for fat. After E 1608 kcal (6.7 MJ) E 1156 kcal (4.8 MJ) E 1302 kcal (5.5 MJ) E 1247 kcal (5.2 MJ)
participants to follow 2 wk, increase CHO CHO 47%, Pro 16% CHO 41 g (15%), CHO 97 g (30%), CHO 53%, Pro 18%
the Atkins diet. to 40 to 60 g/d. Fat 37% (SFA 12.4%) Pro 28%, Fat 57% Pro 23%, Fat 46% Fat 29% (SFA 11%)
(SFA 21%) (SFA 17%)
Truby 2008 A copy of Dr Atkins' Participants were asked Baseline 2 mo Not reported Control 2 mo
UK27 New Diet Revolution to follow the book. E 2283 ± 643 kcal E 1627 ± 551 kcal E 1899 ± 624 kcal
book was provided. Ad libitum (9.6 ± 2.7 MJ) (6.8 ± 2.3 MJ) (8 ± 2.6 MJ)
CHO 42%, Pro 16% CHO 12%, Pro 28%, CHO 42%, Pro 16%,
Fat 37% Fat 57%, Alcohol Fat 37%, Alcohol 5%
3%
Gardner 2010 A copy of Dr Atkins' Induction phase <20 g Baseline 8 wk Not reported LEARN 8 wk
USA26 New Diet Revolution CHO per day, E 1929 ± 509 kcal E 1373 ± 340 kcal E 1478 ± 444 kcal
book was provided. duration depended on (8.1 ± 2.1 MJ) (5.7 ± 1.4 MJ) (6.2 ± 1.9 MJ)
participants. Ongoing CHO 46%, Pro 17% CHO 58 g (17%), CHO 49%, Pro 20%,
phase CHO <50 g/d. Fat 36% (SFA 12%) Pro 28%, Fat 55% Fat 30% (SFA 10%)
(SFA 20%)
Johnston 2011 Research staff supplied Low‐carb, high‐protein: Baselineb N/A 4 wkb N/A
UK25 all foods and drinks 4% CHO, 30% E 2744 kcal (11.5 MJ) E 1906 kcal (8 MJ) CHO
for a very low protein, 66% fat. 70% CHO 52%, Pro 12% 22 g (4%), Pro 29%,
carbohydrate diet. of energy Fat 38% (SFA 13%) Fat 67% (SFA 21%)
requirement.
Bazzano 2014 Dietitian led counselling <40 g CHO per day, no Baseline 3 mo 12 mo LFD 12 mo
USA31 sessions. A handbook specific energy goal. E 1998 ± 740 kcal E 1258 ± 409 kcal E 1448 ± 610 kcal E 1527 ± 522 kcal
of diet recipes, menu, Participants also (8.4 ± 3.1 MJ) (5.3 ± 1.7 MJ) (6.1 ± 2.6 MJ) (6.4 ± 2.2 MJ)
food and shopping received education on CHO 48%, Pro 17% CHO 97 g (29%), CHO 127 g (34%), CHO 54%, Pro 19%,
lists, meal planner SFA, MUFA, and trans Fat 33% (SFA 11%) Pro 26%, Fat 43% Pro 24%, Fat 41% Fat 30% (SFA 9%)
was provided. One fats, with emphasis (SFA 14%) (SFA 13%)
meal replacement per on the benefits of
day was provided. MUFA and to limit 6 mo
trans fats. E 1324 ± 537 kcal
(5.5 ± 2.2 MJ)

(Continues)
1137
TABLE 2 (Continued)
1138

Comparator
Author/Country Mode of Diet Delivery Prespecified CRD Baseline CRD (Intermediate) CRD (Last Time Point) (Last Time Point)
CHO 93 g (28%), Pro
26%, Fat 44% (SFA
13%)
Genoni 2016 A copy of the Paleo Diet Advise for lean meats, Baseline N/A 4 wk AGHE 4 wk
Australia16 book was provided. fish, egg, nuts, fruits E 1864 ± 461 kcal E 1414 ± 347 kcal E 1591 ± 412 kcal
and vegetables, and (7.8 ± 1.9 MJ) (5.9 ± 1.5 MJ) (6.7 ± 1.7 MJ)
almond milk. No corn, CHO 39%, Pro 21% CHO 103 g (28%), CHO 41%, Pro 22%,
white potatoes, and Fat 34% (SFA 12%) Pro 27%, Fat 40% Fat 33% (SFA 12%)
legumes. No grains, (SFA 12%)
cereals, and dairy
products. Ad libitum
Manousou 2018 Participants attended 30% CHO, 30% Pro, Baseline 6 mo 24 mo NNR 24 mo
Sweden17 group session with 40% fat. Advise for E 1993 ± 412 kcal E 1590 ± 327 kcal E 1531 ± 436 kcal E 1716 ± 279 kcal
diet and cooking lean meats, fish, egg, (8.3 ± 1.7 MJ) (6.7 ± 1.4 MJ) (6.4 ± 1.8 MJ) (7.2 ± 1.2 MJ)
practice for a nuts, fruits, and CHO 44%, Pro 17% CHO 29%, Pro 24%, CHO 31%, Pro 23%, CHO 43%, Pro 17%,
Paleolithic diet. vegetables. No Fat 34% Fat 44% Fat 42% Fat 35%
cereals, beans, refined
fat and sugar, salt,
soft drinks and bakery
products. Ad libitum
Mardinoglu Research staffs supplied “Isocaloric” CRD, did not Baseline N/A 2 wk N/A
2018 all foods and drinks aim for weight loss. E 2235 ± 221 kcal E 3115 ± 410 kcal
Sweden28 (9.4 ± 0.9 MJ) (13 ± 1.7 MJ)
CHO 40%, Pro 20% CHO 4%, Pro 24%,
Fat 40% (SFA 14%) Fat 72% (SFA 10%)
Elidottir 2016 Voluntary adherence N/A, cross‐sectional N/A N/A E 1928 ± 534 kcal N/A
Iceland32 to CRD. study (8.1 ± 2.2 MJ)
CHO 10%, Pro 23%,
Fat 66% (SFA 26%)

Abbreviations: AGHE, Australian Guideline to Healthy Eating; CHO, carbohydrate; CRD, carbohydrate‐restricted diet; E, energy; LFD, low fat diet; LEARN, Lifestyle, Exercise, Attitudes, Relationships, Nutrition;
MTV/MTM, multivitamins/multimineral; MUFA, monounsaturated fatty acids; N/A, not applicable. NNR, Nordic Nutrition Recommendations; OWL, ongoing weight loss; Pro, protein; SFA, saturated fatty acids;
SFA, saturated fatty acids.
a
Values are presented as mean ± SD unless otherwise indicated.
b
No SD was reported.
CHURUANGSUK
ET AL.
TABLE 3 Actual intakes of vitamins, percentage of recommended intakes and changes from baselinea

Thiamine Folate B12 Vitamin C

EI EI
Time BW Change, Change, Intake, Baseline Intake, Baseline Baseline Intake, Baseline
CHURUANGSUK

Points Loss, kg kcal MJ mg %RI Change, % μg %RI Change, % Intake, μg %RI Change, % mg %RI Change, %
ET AL.

Miller (M) Baseline 2.8 (2.1) 233 496 (446) 124 8 (6.3) 333 90 (77) 100
2003 US 2 wk −3.8 −1081 −4.5 3.9 (8.7) 325 +39 142 (62) 36 −71 7.6 (6.4) 317 −5 31 (27) 34 −66
4 wk −5.3 −923 −3.9 0.9 (0.3) 75 −68 191 (56) 48 −62 9 (3.3) 375 +13 26 (22) 29 −71
Miller (W) Baseline 1.5 (0.3) 136 209 (65) 52 4.3 (1.3) 179 84 (46) 112
2003 US 2 wk −3.8 −1081 −4.5 0.9 (1.1) 82 −40 236 (279) 59 +13 7 (4.7) 292 +63 48 (48) 64 −43
4 wk −5.3 −923 −3.9 0.8 (0.4) 73 −47 177 (105) 44 −15 5.6 (2.8) 233 +30 41 (33) 55 −51
Brehm Baseline ‐ ‐ ‐ 155.1 39 ‐ ‐ ‐ 70.3 94
2003 US 3 mo −7.6 −452 −1.9 ‐ ‐ ‐ 139.7 35 −10 ‐ ‐ ‐ 35.7 48 −49
6 mo −8.5 −306 −1.3 ‐ ‐ ‐ 195.9 49 +26 ‐ ‐ ‐ 58.5 78 −17
Truby Baseline ‐ 223 (186) 139 (49) ‐ ‐ ‐ 224 (126)
2008 UK 2 mo −5.2 −656 −2.7 ‐ 146 (49) −35 93 (44) −33 ‐ ‐ ‐ 132 (104) −41
Gardner Baseline 1.6 (0.5) 146 535 (207) 134 4.9 (4.3) 204 94 (59) 125
2010 US 2 mo −4.3 −556 −2.3 0.9 (0.4) 82 −44 329 (141) 82 −39 6.6 (7.9) 275 +35 66 (39) 88 −30
Johnston Baseline ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ 153 383
2011 UK 4 wk −6.8 −838 −3.5 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ 148 370 −3
Bazzano Baseline ‐ ‐ ‐ 410 (190) 103 ‐ ‐ ‐ 87 (60) 118
2014 US 3 mo −5.7 −740 −3.1 ‐ ‐ ‐ 290 (130) 73 −29 ‐ ‐ ‐ 68 (45) 90 −24
6 mo −5.6 −674 −2.8 ‐ ‐ ‐ 320 (170) 80 −22 ‐ ‐ ‐ 81 (55) 108 −9
12 mo −5.3 −550 −2.3 ‐ ‐ ‐ 310 (150) 78 −24 ‐ ‐ ‐ 73 (68) 97 −18
Genoni 2016 Baseline 1.55 (0.6) 141 396 (136) 99 ‐ ‐ ‐ 107 (50) 238
Australia 4 wk −3.2 −450 −1.9 0.96 (0.5) 87 −38 317 (85) 79 −20 ‐ ‐ ‐ 168 (84) 373 +57
Mardinoglu Baseline 1.64 (0.4) 126 365 (168) 122 11 (12) 567 142 (83) 190
2018 Sweden 2 wk −1.9 +880 +3.7 2.9 (0.5) 223 +77 345 (16) 115 −5 12 (0.7) 607 +7 154 (6) 206 +8
Elidottirb,c Men ‐ ‐ ‐ 1.6 (1.4) 123 ‐ 319 (295) 106 ‐ 7 (5) 350 ‐ 53 (77) 71 ‐
2016 Iceland Women ‐ ‐ ‐ 1.0 (1.3) 91 ‐ 275 (335) 92 ‐ 7 (4) 350 ‐ 76 (92) 101 ‐

Vitamin A β‐Carotene Vitamin E

BW Loss, EI Change, EI Change, Baseline Baseline Baseline

Time Points kg kcal MJ Intake, μg %RI Change, % Intake, μg %RI Change, % Intake, mg %RI Change, %

Miller (M) 2003 US Baseline 1553 (1733) 173 ‐ ‐ ‐ 15 (16) 100

2 wk −3.8 −1081 −4.5 513 (238 57 −67 ‐ ‐ ‐ 29 (69) 193 +93
4 wk −5.3 −923 −3.9 622 (355 69 −60 ‐ ‐ ‐ 7.1 (2.5) 47 −53
Miller (W) 2003 US Baseline 779 (387) 111 ‐ ‐ ‐ 9.4 (3.8) 63
2 wk −3.8 −1081 −4.5 1023 (1094) 146 +31 ‐ ‐ ‐ 11 (15) 71 +14
4 wk −5.3 −923 −3.9 703 (385) 100 −10 ‐ ‐ ‐ 7 (5) 47 −26
1139

(Continues)
TABLE 3 (Continued)
1140

Vitamin A β‐Carotene Vitamin E

BW Loss, EI Change, EI Change, Baseline Baseline Baseline

Time Points kg kcal MJ Intake, μg %RI Change, % Intake, μg %RI Change, % Intake, mg %RI Change, %
Brehm 2003 US Baseline ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐
3 mo −7.6 −452 −1.9 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐
6 mo −8.5 −306 −1.3 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐
Truby 2008 UK Baseline ‐ 182 (192) ‐ ‐ ‐ ‐ ‐ ‐
2 mo −5.2 −656 −2.7 ‐ 175 (257) −4 ‐ ‐ ‐ ‐ ‐ ‐
Gardner 2010 US Baseline 681 (289) 97 ‐ ‐ ‐ 8.4 (4.1) 56
2 mo −4.3 −556 −2.3 743 (323) 106 +9 ‐ ‐ ‐ 8.7 (4.8) 58 +4
Johnston 2011 UK Baseline 533 76 3705 ‐ 7.2 n/a
4 wk −6.8 −838 −3.5 625 89 +17 3778 ‐ +2 11.9 n/a +65
Bazzanoc 2014 US Baseline ‐ ‐ ‐ 490 (1310) ‐ ‐ ‐ ‐
3 mo −5.7 −740 −3.1 ‐ ‐ ‐ 890 (1990) ‐ +82 ‐ ‐ ‐
6 mo −5.6 −674 −2.8 ‐ ‐ ‐ 600 (1500) ‐ +22 ‐ ‐ ‐
12 mo −5.3 −550 −2.3 ‐ ‐ ‐ 420 (1420) ‐ −14 ‐ ‐ ‐
Genoni 2016 Australia Baseline 1015 (448) 145 4258 (2508) ‐ 9.8 (2.5) 140
4 wk −3.2 −450 −1.9 2032 (1359) 290 +100 11 379 (7966) ‐ +167 14.3 (7.3) 204 +46
Mardinoglu 2018 Sweden Baseline 1517 (2243) 169 4116 (2503) ‐ 13.8 (4.5) 138
2 wk −1.9 +880 +3.7 724 (35) 80 −52 3011 (134) ‐ −27 30.9 (1.5) 309 +125
Elidottirb,c 2016 Iceland Men ‐ ‐ ‐ 956 (773) 106 ‐ ‐ ‐ ‐ 16 (13) 160 ‐
Women ‐ ‐ ‐ 880 (472) 126 ‐ ‐ ‐ ‐ 17 (8) 213 ‐

Abbreviations: BW, body weight; EI, energy intake; M, men; RI, recommended intake; W, women.
a
Values are expressed as mean (SD) unless otherwise indicated.
b
Cross‐sectional study.
c
Values are median (interquartile range).
CHURUANGSUK
ET AL.
TABLE 4 Mineral intakes, percentage of recommended intakes, and changes from baselinea

Calcium Magnesium Iron

Baseline
CHURUANGSUK

Time BW EI Change, EI Change, Change, Baseline Baseline

Points Loss, kg kcal MJ Intake, mg %RI % Intake, mg %RI Change, % Intake, mg %RI Change, %
ET AL.

Miller (M) 2003 US Baseline 1139 (458) 114 350 (149) 83 21 (12) 263
2 wk −3.8 −1081 −4.5 420 (180) 42 −63 163 (61) 39 −53 9 (4) 113 −57
4 wk −5.3 −923 −3.9 503 (153) 50 −56 180 (41) 43 −49 12 (3) 150 −43
Miller (W) 2003 US Baseline 801 (309) 80 280 (72) 88 13 (3) 72
2 wk −3.8 −1081 −4.5 513 (223) 51 −36 129 (42) 40 −54 8 (2) 44 −39
4 wk −5.3 −923 −3.9 517 (268) 52 −36 182 (163) 57 −35 8 (4) 44 −39
Brehm 2003 US Baseline 591 59 ‐ ‐ ‐ ‐ ‐ ‐
3 mo −7.6 −452 −1.9 444 44 −25 ‐ ‐ ‐ ‐ ‐ ‐
6 mo −8.5 −306 −1.3 739 74 +25 ‐ ‐ ‐ ‐ ‐ ‐
Truby 2008 UK Baseline 125 (44) ‐ 106 (33) ‐ 136 (60)
2 mo −5.2 −656 −2.7 92 (38) −26 ‐ 75 (38) −29 ‐ 91 (49) −33
Gardner 2010 US Baseline 851 (314) 85 291 (88) 91 14.8 (4.8) 82
2 mo −4.3 −556 −2.3 742 (273) 74 −13 231(86) 72 −21 10.5 (4.1) 58 −29
Genoni 2016 Baseline 771 (204) 77 359 (85) 112 12.7 (2.6) 71
Australia 4 wk −3.2 −450 −1.9 355 (91) 36 −54 328 (97) 103 −9 11.1 (2.7) 62 −13
Mardinoglu 2018 Baseline 1187 (342) 148 414 (74) 118 15.7 (3.3) 175
Sweden 2 wk −1.9 +880 +3.7 1160 (472) 145 −2 338 (49) 97 −18 15.7 (1.6) 171 −2
b,c
Elidottir 2016 Men ‐ ‐ ‐ 770 (564) 96 ‐ 245 (148) 70 ‐ 11 (7) 122 ‐
Iceland Women ‐ ‐ ‐ 774 (566) 97 ‐ 254 (114) 91 ‐ 9 (6) 60 ‐

Iodine Zinc Selenium

BW Loss, EI Change, EI Change, Baseline Baseline Baseline
Time Points kg kcal MJ Intake, μg %RI Change, % Intake, μg %RI Change, % Intake, μg %RI Change, %

Miller (M) 2003 US Baseline ‐ ‐ ‐ 16 (7) 146 167 (68) 304

2 wk −3.8 −1081 −4.5 ‐ ‐ ‐ 14 (5) 127 −13 142 (107) 258 −15
4 wk −5.3 −923 −3.9 ‐ ‐ ‐ 17 (4) 155 +6 146 (54) 266 −13
Miller (W) 2003 US Baseline ‐ ‐ ‐ 10 (2) 125 128 (21) 233
2 wk −3.8 −1081 −4.5 ‐ ‐ ‐ 11 (4) 138 +10 108 (34) 196 −16
4 wk −5.3 −923 −3.9 ‐ ‐ ‐ 12 (6) 150 +20 112 (40) 204 −13
Brehm 2003 US Baseline ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐
3 mo −7.6 −452 −1.9 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐
6 mo −8.5 −306 −1.3 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐
Truby 2008 UK Baseline ‐ ‐ ‐ ‐ 119 (33) ‐ 100 (33)
2 mo −5.2 −656 −2.7 ‐ ‐ ‐ ‐ 139 (49) +17 ‐ 149 (225) +49

(Continues)
1141
 1142 CHURUANGSUK ET AL.

baseline, but the “Paleolithic diet” showed a significant increase

Change, %
Baseline
in vitamin C intake by 57%. Similarly, Atkins diet trials (n = 6)
showed inconsistent findings regarding vitamin A, E, and β‐carotene

+6.5

+63
‐
‐

‐
‐
intakes, while a single “Paleolithic diet” trial showed increases in these
micronutrients (vitamin A, E, and β‐carotene intakes) by +100%, +46%,
and +167%, respectively (Table 3). Intakes of zinc and selenium did
%RI
195
207

149
244
not change markedly; the intakes at baseline and at the end of
‐
‐

‐
‐
trials were both above recommended intakes (Table 4). Dietary

89.7 (34.4)
146.5 (8.8)
data for vitamin B12, A, C, E, β‐carotene, zinc, and selenium among
107 (36)
114 (44)
Intake, μg
Selenium

‐
‐

‐
‐
the included studies were assessed by either self‐reported
dietary records, 24‐hour recalls, or analysis of all foods provided by
researchers.
Change, %
Baseline

+11

3.5 | Correlations between changes in body weight

+3

+3

‐
‐

and micronutrient intakes

There were strong correlations between amount of weight loss (kg)

%RI
134
138
144
148
153
170
‐
‐

and actual intakes of thiamine (rho = 0.83, P = 0.167), magnesium

(rho = 0.95, P = 0.051), and iron (rho = 0.95, P = 0.051). As percentage
Intake, μg
10.7 (3.7)
11 (3.5)
11.5 (2.4)
11.8 (3.4)
13.8 (2.3)
15.3 (1)

of Reference Nutrient Intake (%RNI), there were only strong correla-

‐
‐
Zinc

tions between weight loss and %RNI of magnesium (rho = 0.92,

P = 0.026) and iron (rho = 0.82, P = 0.089).
Change, %
Baseline

3.6 | Micronutrient adequacy

−47

−37
‐
‐

‐
‐

No case of clinical micronutrient deficiency was reported. Only one

%RI

187
118
80
43
‐
‐

‐
‐

RCT by Gardner et al reported micronutrient adequacy by showing

the proportion of participants whose intakes were below estimated
Abbreviations: BW, body weight; EI, energy intake; M, men; RI, recommended intake; W, women.
Intake, μg

281 (133)
120 (39)
64 (23)

177 (10)

average requirement (EAR) after 8 weeks on an Atkins diet for weight

Iodine

‐
‐

‐
‐

loss. Three nonconsecutive 24‐hour dietary recalls were used in this

study. Intakes of thiamine, folate, and vitamin C all fell below EAR in
EI Change,

approximately 50% of participants, 70% for magnesium, and 30% for

iron. Importantly, there was a threefold increase in the proportion of
−2.3

−1.9

+3.7
MJ

participants who had iron intakes below EAR at the end of the Atkins
‐
‐

diet trial (from 3% to 32% of participants).

EI Change,

Values are expressed as mean (SD) unless otherwise indicated.

−556

−450

+880
kcal

3.7 | Micronutrient status

‐
‐
BW Loss,

Measure of micronutrient status were reported only in four studies,

including Atkins diet (n = 1), Paleolithic diet (n = 2), and “isocaloric”
−4.3

−3.2

−1.9
kg

CHO‐restricted diet (n = 1) for folate, vitamin A, carotenoids, vitamin

‐
‐

E, and iodine (Table 6). Serum and red blood cell (RBC) folate
Values are median (Interquartile range).
Time Points

increased after 2 weeks of “isocaloric” CHO‐restricted diet

Baseline

Baseline

Baseline

Women

(+9 nmol/L) and 4 weeks of a “Paleolithic” diet (+90 nmol/L), respec-

2 mo

4 wk

2 wk
Men

tively, which are in disagreement with reported folate intakes

(Table 3). The Atkins diet resulted in an increase in plasma vitamin
Genoni 2016 Australia

Cross‐sectional study

C concentration but reductions (−2 to −40% from baseline) in plasma

(Continued)

Gardner 2010 US

Mardinoglu 2018

retinol, carotenoids, and α‐tocopherol. In contrast, the “Paleolithic”

2016 Iceland

diet reported increases in plasma lycopene, α‐carotenes, and

Sweden
Elidottirb,c

β‐carotenes (Table 6).

TABLE 4

Iodine status was reported in only one RCT showing a significant

reduction in median 24‐hour urine iodine concentration after a
b
a

c
CHURUANGSUK ET AL. 1143

TABLE 5 Micronutrient intakes following carbohydrate‐restricted diets compared with control dietsa

Brehm (2003) Truby (2008) Gardner (2010)

3 mob 6 mob 2 moc 2 mod

%RNI
Atkins LFD Atkins LFD Atkins Control Atkins LEARN
Micronutrients (n = 22) (n = 20) (n = 22) (n = 20) (n = 30) (n = 26) (n = 73) (n = 73)

Thiamine, mg ‐ ‐ ‐ ‐ 146 (49) 178 (76) 0.9 (0.4) 1.4 (0.4)*

Folate, μg 139.7 221.7 195.9 193.9 93 (44) 114 (36) 329 (141) 470 (190)***
Vitamin B12, μg ‐ ‐ ‐ ‐ ‐ ‐ 6.6 (7.9) 5.7 (10.2)
Vitamin C, mg 35.7 94.2** 58.5 53.1 132 (104) 205 (214) 66 (39) 100 (62)***
Vitamin A, μg ‐ ‐ ‐ ‐ 175 (257) 118 (66) 743 (323) 689 (327)
Carotene, μg ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐
Vitamin E, mg ‐ ‐ ‐ ‐ ‐ ‐ 8.7 (4.8) 7.2 (4.2)
Magnesium, mg ‐ ‐ ‐ ‐ 75 (38) 95 (31) 231 (86) 286 (89)**
Calcium, mg 444.2 567.2 739.0 662.6 92 (38) 111 (41) 742 (273) 801 (339)
Iron, mg ‐ ‐ ‐ ‐ 91 (49) 104 (46) 10.5 (4.1) 13.7 (5.2)***
Zinc, μg ‐ ‐ ‐ ‐ 139 (49) 103 (31) 11 (3.5) 8.9 (3.1)***
Selenium, μg ‐ ‐ ‐ ‐ 149 (225) 87 (61) 114 (44) 97 (34)***
Iodine, μg ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐
Bazzano (2014) Genoni (2016)
3 moc 6 moc 12 moc 4 wke
CRD LFD CRD LFD CRD LFD Paleolithic diet AGHE
Micronutrients n = 62 n = 61 n = 54 n = 50 n = 54 n = 49 (n = 22) (n = 17)

Thiamine, mg ‐ ‐ ‐ ‐ ‐ ‐ 0.96 (0.5) 1.49 (11)*

Folate, μg 290 (130) 360 (230) 320 (170) 380 (220) 310 (150) 350 (200) 317(85) 350 (132)
Vitamin B12, μg ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐
Vitamin C, mg 68 (45) 83 (68) 81 (55) 85 (66) 73 (68) 83 (61) 168 (84) 95.6 (51)*
Vitamin A, μg ‐ ‐ ‐ ‐ ‐ ‐ 2032 (1359) 984 (679)*
Carotene, μg 890 (1990) 990 (2030) 600 (1500) 740 (2440) 420 (1420) 1000 (2350) 11 379 (7966) 4386 (3642)**
Vitamin E, mg ‐ ‐ ‐ ‐ ‐ ‐ 14.3 (7.3) 9 (5)**
Magnesium, mg ‐ ‐ ‐ ‐ ‐ ‐ 328 (97) 366 (106)
Calcium, mg ‐ ‐ ‐ ‐ ‐ ‐ 355 (91) 759 (247)**
Iron, mg ‐ ‐ ‐ ‐ ‐ ‐ 11.1 (2.7) 10.6 (3.2)
Zinc, μg ‐ ‐ ‐ ‐ ‐ ‐ 11.8 (3.4) 10.3 (3.1)
Selenium, μg ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐
Iodine, μg ‐ ‐ ‐ ‐ ‐ ‐ 64 (23) 145 (40)**

Abbreviations: AGHE, Australian Guideline Healthy Eating; CRD, carbohydrate‐restricted diet; LFD, low‐fat diet; RNI, reference nutrient intake.
a
Values are mean (SD) unless otherwise indicated.
b
No SD reported.
c
No statistical comparison between groups.
d
P values obtained from ANOVA, pairwise comparisons were significantly different (Tukey's Studentized range test).
e
P values for mean difference of change (end‐baseline) between groups.
*P < 0.05.
**P < 0.01.
***P < 0.001 compared between groups.
1144 CHURUANGSUK ET AL.

TABLE 6 Plasma and urine biomarkers of micronutrientsa 4 | DISCUSSION

Biochemical Measures Baselineb End P value
Most CHO‐restricted diets are used for weight control. Losing weight
b
Atkins diet 4 wk
requires an energy‐restricted (ie, macronutrient‐restricted) diet for a
(Johnstone 2011)
significant period of time. However, the body still requires adequate
Retinol, μg/mL 0.623 0.456 <0.001
micronutrient intake to maintain normal metabolic function. If the
α‐Tocopherol, μg/mL 11.46 9.30 <0.001 weight loss diet is not well designed nutritionally, micronutrient
β‐Cryptoxanthin, μg/ 0.065 0.039 <0.001 insufficiencies can develop. The present paper is a systematic review
mL
of published evidence, up to October 2018, on the impact of CHO‐
Lycopene, μg/mL 0.392 0.244 <0.001 restricted diets on micronutrient intakes and status. Although rather
α‐Carotene, μg/mL 0.039 0.035 0.002 few published studies on CHO‐restricted diets have reported their
β‐Carotene, μg/mL 0.135 0.132 0.004 effects on micronutrients, we have identified some consistent patterns
Xanthophyll, μg/mL 0.186 0.207 <0.001 of reduced micronutrient intakes, and to levels lower than control‐
Vitamin C, μmol/L 46.5 51.7 <0.001 diets, when people follow diets which limit CHO‐rich foods. A variety
of CHO‐restricted diets were examined, mostly primarily aiming for
Paleolithic diet 4 wk
(Genoni 2016) weight control. Given the lack of consistent difference in weight loss,

Lycopene, μmol/L 0.61 ± 0.3 0.52 ± 0.2 ns compared with higher CHO diets,6,13,14 these findings might provide
grounds to prefer weight‐control diets which do not excessively
α‐Carotene, μmol/L 0.39 ± 0.4 0.46 ± 0.2 ns
restrict CHO more than other macronutrients.
β‐Carotene, μmol/L 1.54 ± 1.3 2.41 ± 1.8 <0.05
Intakes of thiamine, folate, magnesium, calcium, iron, and iodine all
Red cell folate, nmol/L 894 ± 214 981 ± 300 <0.01
decreased after following any types of CHO‐restricted diet (Figure 2).
“Isocaloric” CRD 2 wk This probably reflects the fact that CHO‐rich staple foods are a good
(Mardinoglu 2018)
source of vitamins and minerals, either naturally presented or being
Serum folate, nmol/L 25.6 ± 4.1 34.5 ± 5.5 <0.001
fortified, and for some, they usually form the main sources to reach
Paleolithic diet 6 moc End (2 y)c dietary reference values (DRVs). For example, the DRV (United
(Manousou 2018)
Kingdom) of thiamine for a healthy adult man is 1 mg/d. This would
24‐h UIC, μg/L 68 36 (IQR 36), 57
be met by about 100 g of commercial breakfast cereal, 400 g of
(IQR 37) P = 0.001 (IQR 32),
P = 0.033
wholemeal breads. Not all CHO‐rich foods are such rich source of
thiamine, thus to reach the DRV would require about 500 g of baked
24‐h UIE, μg/L 124.5 (IQR 77 (IQR 65), 113.5
88) P = 0.001 (IQR 79), ns potato, but this compares with 1000 g of beef, or 1250 g of boiled
eggs, or 5000 g of cheese.15 Although iodine is not a nutrient acquired
Proportion of subjects Baseline 6 mo (n = 30) P valued
according to 24‐h UIC, (n = 34) from CHO‐rich foods, both Atkins and “Paleolithic” diets exclude
n (%) milk/dairy from the diets leading to lower iodine intake. Our system-
<50 μg/L 5 (15) 22 (73) 0.001 atic review showed that both iodine intake and iodine status (urinary
51 to 100 μg/L 23 (68) 5 (17) 0.001 biomarker) were reduced after following a “Paleolithic” diet.16,17

101 to 200 μg/L 6 (18) 3 (10) Ns

In the United Kingdom, bread and fortified breakfast cereals, milk
and milk products, and vegetable including potatoes are the top three
>200 μg/L 0 0
food groups as sources of the vitamins and minerals,18 which are
Abbreviations: CRD, carbohydrate‐restricted diet; IQR, interquartile range; reported to be reduced on CHO‐restricted diets.
ns, not significant; UIC, urine iodine concentration; UIE, urine iodine
The reduced intakes with CHO‐restricted diets in studies reviewed
excretion.
a here could have important clinical consequences. There have been
Values are mean ± SD unless otherwise indicated.
b
case reports of severe, potentially life‐threatening, thiamine deficiency
No SD reported.
c
from following extreme or prolonged CHO‐restricted diets. Bilateral
P value compared with baseline.
d
optic neuropathy was reported in two patients who followed a
P value from the longitudinal analysis compared with baseline.
prolonged CHO‐restricted diet,19 and Wernike's encephalopathy
with cardiac beriberi was reported in a young man with obesity who
“Paleolithic” diet at both 6 months (36 μg/L) and 2 years (57 μg/L), cut out all breads, cereal, and pasta for several months.20 Given that
compared with baseline (68 μg/L), one‐fold to two‐fold lower than weight loss may promote reproductive function in women with
the control diet (no change through the study, 70 μg/L) (Table 6). overweight/obesity, inadequate intake of folate, and iodine in women
Importantly, there was a rise in proportion of women with urine iodine of child‐bearing age both increase risk of fetal maldevelopment, and
concentration less than 50 μg/L, which is the internationally agreed iron deficiency is already frequent in young women.21,22 Increased
cut‐off for moderate iodine deficiency, from 15% at baseline to 73% risk by 17% to 22% of type 2 diabetes has been reported with low
at 6 months on the “Paleolithic” diet (Table 6). magnesium consumption.23,24
CHURUANGSUK ET AL. 1145

FIGURE 2 Percentage of changes from baseline intakes at the end of each study by micronutrients [Colour figure can be viewed at
wileyonlinelibrary.com]

For other vitamins, different advice between specific CHO‐ potential for serious deficiencies, and because there are always
restricted diets regarding fruits and vegetables intake probably questions about the reliability of dietary assessment methods to
explains the variable effects on some micronutrients. Reductions in report intakes. Two out of 10 included studies used the memory‐
carotenoid intakes and plasma concentrations were reported after an based 24‐hour dietary recall, while the rest used a prospective
Atkins diet,25 while there were increases in carotenoid intakes and approach (dietary record) and/or supplied all foods and drinks to
plasma concentrations after a “Paleolithic” diet.16 The “Paleolithic” participants. Only four out of 10 studies employed an objective
diet16 also showed an increase in vitamin C intake, reflecting advice biomarker measurement for at least one micronutrient.
to consume more fruits than typical Atkins diets, which mostly There is a wide variety of what people consider to be a “low‐CHO
showed reductions in vitamin C intakes26,27 (Figure 2). However, one diet,” as low as 20 to 50 g of CHO per day or up to 40% to 45% of
trial with a modified Atkins diet showed an increase in plasma vitamin total energy intake as seen in many low‐CHO diet papers.6,7,16,26
C concentration, probably because fruit juice was included in the For people who are not wishing to lose weight, under 50 g of CHO
25
meals provided. per day leaves a large (somewhat less practical) amount of calories
Despite reduced intake of folate, serum and RBC folate increased to be derived from non‐CHO sources. The compositions of control
following an “isocaloric” CHO‐restricted diet and one “Paleolithic” diets in RCTs also varied in macronutrient contents.
diet. This is proposed to be due to a CHO‐restricted diet induced Only one trial reported proportion of participants below EAR, but
increment in folate‐producing gut bacteria as demonstrated in the this could be more clinically meaningful if all studies could have
study by Mardinoglu et al,28 or it may reflect a limitation in the food reported on proportion below lower reference nutrient intakes.
composition database used.16 The intakes of some micronutrients Another important point relates to the between‐country variation in
(status not reported), specifically B12, zinc, and selenium, did terms of recommended intake, which would lead to different apparent
consistently increase after CHO‐restricted diet—however, intakes dietary adequacy for some micronutrients. For example, the UK RNI
were adequate at baseline. for folate and vitamin C are 200 mcg and 40 mg per day while these
are 400 mcg and 90 mg for the United States. If the US recommended
intakes were used for all studies, the numbers of inadequate intakes
4.1 | Limitations and strength reported could be greater in some other countries.
Despite the Atkins diet book recommending dietary supplements,
It was disappointing, but perhaps interesting given the amount of most of the included studies did not report on supplement use causing
media interest, that so few of the many studies on CHO‐restricted difficulties in estimating of how significant the problem of inadequacy
diets have reported micronutrient intakes or status. The lack of is. There is no guideline on the use of micronutrient supplementation
reporting of measured micronutrient status is concerning, given the while on energy restricted diets. The present systematic review has
1146 CHURUANGSUK ET AL.

found frequent and consistent reductions in intake or status, such that control in people with type 2 diabetes: a systematic review including
micronutrient supplementation might be recommended routinely with GRADE assessments. Am J Clin Nutr. 2018;108(2):300‐331.

CHO‐restricted diets. 8. Gardner CD, Trepanowski JF, del Gobbo LC, et al. Effect of low‐fat vs
low‐carbohydrate diet on 12‐month weight loss in overweight adults
and the association with genotype pattern or insulin secretion: the
5 | C O N CL U S I O N DIETFITS randomized clinical trial. JAMA. 2018;319(7):667‐679.
9. American Diabetes A. 4. Lifestyle management: standards of medical
care in diabetes‐2018. Diabetes Care. 2018;41(Supplement 1):S38‐S50.
This study has used rigorous systematic review methodology to estab-
lish a rather consistent finding of multiple micronutrient inadequacy in 10. Jensen MD, Ryan DH, Apovian CM, et al. 2013 AHA/ACC/TOS guide-
line for the management of overweight and obesity in adults: a report
free‐living participants following CHO‐restricted diet. All included
of the American College of Cardiology/American Heart Association
studies seem to be at low risk of bias. Future research should report Task Force on Practice Guidelines and The Obesity Society. Circulation.
on micronutrient status as well as dietary intakes, to inform not only 2014;129(25 suppl 2):S102‐S138.
the benefits of the intervention, but also the risk of micronutrient 11. Collins CB, Winham DM, Hutchins AM, Salbe AD. Dietary intake, char-
inadequacies, which could help readers/practitioners' decision making. acteristics, and attitudes of self‐reported low‐carbohydrate dieters.
FASEB J. 2006;20:A575‐A575.
There is a case for recommending routine use of micronutrient
12. Moher D, Liberati A, Tetzlaff J, Altman DG, Group Prisma. Preferred
supplementation with CHO‐restricted diets.
reporting items for systematic reviews and meta‐analyses: the PRISMA
statement. BMJ. 2009;339(jul21 1):b2535.
ACKNOWLEDGEMEN T
13. Johnston BC, Kanters S, Bandayrel K, et al. Comparison of weight
C.C. was funded for his PhD scholarship by the Prince of Songkla loss among named diet programs in overweight and obese adults: a
University, Faculty of Medicine, Thailand. No other external funding meta‐analysis. JAMA. 2014;312(9):923‐933.

was received. 14. Mansoor N, Vinknes KJ, Veierod MB, Retterstol K. Effects of low‐
carbohydrate diets v. low‐fat diets on body weight and cardiovascular
risk factors: a meta‐analysis of randomised controlled trials. Brit J Nutr.
CONF LICT OF INT E RE ST 2016;115(03):466‐479.
C.C. has received PhD scholarship from the Prince of Songkla Univer- 15. Public Health England. McCance and Widdowson's 'composition of
sity, Faculty of Medicine, Thailand. M.E.J.L. has received departmental foods integrated dataset' on the nutrient content of the UK food
supply. 2015.
research support from Diabetes UK, Cambridge Weight Plan and
16. Genoni A, Lyons‐Wall P, Lo J, Devine A. Cardiovascular, metabolic
Novo Nordisk and consultancy fees and support for meeting atten-
effects and dietary composition of ad‐libitum Paleolithic vs. Australian
dance from Novo‐Nordisk, Eli Lilly, Novartis, Counterweight Ltd and
guide to healthy eating diets: a 4‐week randomised trial. Nutrients.
Eat Balanced. E.C. and D.G. declare no conflict of interest. 2016;8(5):314.
17. Manousou S, Stal M, Larsson C, et al. A Paleolithic‐type diet results in
ORCID iodine deficiency: a 2‐year randomized trial in postmenopausal obese
women. Eur J Clin Nutr. 2018;72(1):124‐129.
Chaitong Churuangsuk https://orcid.org/0000-0003-4373-6395
18. Bates B., Lennox A., Prentice A., Bates Chris, Page Polly, Nicholson
Michael E.J. Lean https://orcid.org/0000-0003-2216-0083
Sonja, Milne Anne and Swan Gillian National Diet and Nutrition Survey
Emilie Combet https://orcid.org/0000-0002-9302-8971 Results from Years 1, 2, 3 and 4 (combined) of the Rolling Progr`amme
(2008/2009 – 2011/2012). 2014.
RE FE R ENC E S 19. Hoyt CS 3rd, Billson FA. Low‐carbohydrate diet optic neuropathy. Med
1. Finney Rutten LJ, Lazarus Yaroch A, Colon‐Ramos U, Uriyoan AA. J Aust. 1977;1(3):65‐66.
Awareness, use, and perceptions of low‐carbohydrate diets. Prev 20. McKenna LA, Drummond RS, Drummond S, Talwar D, Lean ME. Seeing
Chronic Dis. 2008;5:A130. double: the low carb diet. BMJ. 2013;346(may01 2):f2563.

2. Three million follow Atkins diet. BBC News: 2003. 21. Combet E, Buckton C. Micronutrient deficiencies, vitamin pills and
nutritional supplements. Medicine. 2015;43(2):66‐72.
3. Jallinoja P, Niva M, Helakorpi S, Kahma N. Food choices, perceptions of
22. Gernand AD, Schulze KJ, Stewart CP, West KP Jr, Christian P. Micro-
healthiness, and eating motives of self‐identified followers of a low‐
nutrient deficiencies in pregnancy worldwide: health effects and
carbohydrate diet. Food Nutr Res. 2014;58(1):23552.
prevention. Nat Rev Endocrinol. 2016;12(5):274‐289.
4. Feinman RD, Vernon MC, Westman EC. Low carbohydrate diets in 23. Fang X, Han H, Li M, et al. Dose‐response relationship between dietary
family practice: what can we learn from an internet‐based support magnesium intake and risk of type 2 diabetes mellitus: a systematic
group. Nutr J. 2006;5(1):26. review and meta‐regression analysis of prospective cohort studies.
Nutrients. 2016;8(11).
5. Lean MEJ, Astrup A, Roberts SB. Making progress on the global crisis
of obesity and weight management. BMJ. 2018;361:k2538. 24. Dong J‐Y, Xun P, He K, Qin L‐Q. Magnesium intake and risk of type 2
diabetes: meta‐analysis of prospective cohort studies. Diabetes Care.
6. Churuangsuk C, Kherouf M, Combet E, Lean M. Low‐carbohydrate 2011;34(9):2116‐2122.
diets for overweight and obesity: a systematic review of the systematic
25. Johnstone AM, Lobley GE, Horgan GW, et al. Effects of a high‐protein,
reviews. Obes Rev. 2018;19(12):1700‐1718.
low‐carbohydrate v. high‐protein, moderate‐carbohydrate weight‐loss
7. van Zuuren EJ, Fedorowicz Z, Kuijpers T, Pijl H. Effects of low‐ diet on antioxidant status, endothelial markers and plasma indices of
carbohydrate‐ compared with low‐fat‐diet interventions on metabolic the cardiometabolic profile. Br J Nutr. 2011;106(02):282‐291.
CHURUANGSUK ET AL. 1147

26. Gardner CD, Kim S, Bersamin A, et al. Micronutrient quality of weight‐ 32. Elidottir AS, Halldorsson TI, Gunnarsdottir I, Ramel A. Dietary intake
loss diets that focus on macronutrients: results from the A TO Z study. and cardiovascular risk factors in Icelanders following voluntarily a
Am J Clin Nutr. 2010;92(2):304‐312. low carbohydrate diet. PLoS ONE. 2016;11(8):e0156655.
27. Truby H, Hiscutt R, Herriot AM, et al. Commercial weight loss diets
meet nutrient requirements in free living adults over 8 weeks: a SUPPORTING INFORMATION
randomised controlled weight loss trial. Nutr J. 2008;7(1):25.
Additional supporting information may be found online in the
28. Mardinoglu A, Wu H, Bjornson E, et al. An integrated understanding of
the rapid metabolic benefits of a carbohydrate‐restricted diet on Supporting Information section at the end of the article.
hepatic steatosis in humans. Cell Metab. 2018;27(3):559‐71 e5.
29. Miller BV, Bertino JS, Reed RG, et al. An evaluation of the atkins' diet.
Metab Syndr Relat Disord. 2003;1(4):299‐309.
How to cite this article: Churuangsuk C, Griffiths D, Lean
30. Brehm BJ, Seeley RJ, Daniels SR, D'Alessio DA. A randomized trial
MEJ, Combet E. Impacts of carbohydrate‐restricted diets on
comparing a very low carbohydrate diet and a calorie‐restricted low
fat diet on body weight and cardiovascular risk factors in healthy micronutrient intakes and status: A systematic review. Obesity
women. J Clin Endocrinol Metab. 2003;88(4):1617‐1623. Reviews. 2019;20:1132–1147. https://doi.org/10.1111/
31. Bazzano LA, Hu T, Reynolds K, et al. Effects of low‐carbohydrate and obr.12857
low‐fat diets: a randomized trial. Ann Intern Med. 2014;161(5):
309‐318.